Disclaimer: This interview is for informational purposes only and should not be relied upon as a basis for investment decisions. In Practise is an independent publisher and all opinions expressed by guests are solely their own opinions and do not reflect the opinion of In Practise.

{audio:0:01:10} I am excited to talk to you and learn more about the cloud and Snowflake.

{audio:0:01:24} Is the intent of your research for investment purposes?

{audio:0:01:33} I am a one-person investment shop and I manage money for several wealthy families. I am a very long-term investor and focus on high quality companies that you can ideally own for five to 10 years.I have been intrigued by what Snowflake have achieved and the people they have on board. The purpose of the call today is to talk about the cloud market where you have a lot of experience, and the role the hyperscalers and Snowflake are playing there. 

{audio:0:02:42} It's an interesting time for me and to be totally candid, I left Google to take care of my family. Snowflake, AWS, Workday and several smaller startups reached out to see if I wanted to join them to help lead sales. That's good because I got to engage directly with these companies whereas I used to work with them as a partner or competitor.

{audio:0:03:38} From your CV, I see you worked at the best sales organizations on earth, so it will be great if you could explain the differences between these organizations.

{audio:0:04:00} How much do you know about the companies and their products offerings? Do you have a macro view or do you know details about their strategies?

{audio:0:04:14} I'm not a developer or user of the product, but I've done extensive research and know many of their products, and if I don't know something I will ask. Your company positions were at Microsoft, Salesforce and Google so you've seen them all, could you run through your career, what you learned from each position and what the soul or character of each company was?

{audio:0:05:10} At Microsoft, I led Azure from the beta stages where we did $200,000 all the way to $11 million, so it was like operating the world's largest startup within Microsoft. Microsoft's strength is in the enterprise and their bread and butter at that time was Windows and Office. Once Satya came on board, Microsoft Azure is their highest growing business. Recently Satya told them to slow down hiring on Windows and Office and focus more on Azure. One key distinction I want people to understand is the difference between SaaS, PaaS and IaaS. The street is more focused on IaaS than SaaS applications like Salesforce or Workday because IaaS is unlimited in growth whereas SaaS is bound by the number of users. For example, when I was at Salesforce, I handled one of the largest global energy companies, and as big as they were, the number of users who could use the Salesforce CRM system was capped.

{audio:0:07:10} I guess the market is computing spend.

{audio:0:07:15} Computing spend because apps will always be developed. We are still in the early stages, and 80% to 90% of workloads still need to be migrated as well as all the future workloads.

{audio:0:07:36} Where would you put that number, as AWS claim 5% to 15% is in the cloud? 

{audio:0:07:44} 80% to 90% of the total addressable market can be converted/migrated to cloud.

{audio:0:07:56} Will 100% be in the cloud? 

{audio:0:08:00} Eventually there will be no more data centers, just like the internal combustion engine, where everything will eventually be electric. That could take 20 to 30 years, but during our lifetime, data centers will disappear. Some might remain for specific military operations, but even for those, AWS and Microsoft have outposts and rugged devices which can be put on the edge. Companies will no longer need to invest in growing their data centers.

{audio:0:08:53} I share that view of the world. 

{audio:0:08:59} I got to see Azure from a beta stage to the product it has become. We started out as a PaaS application because Steve Ballmer was into developers and he got part of it right. If you focus on developers, ISVs and applications, that will grow the Azure footprint faster than migrating an application from a company. His mentality was that the more applications built on Azure, the faster it will grow. One of my duties at the time was to find key applications like SAP, then put them on Azure. All my experiences at Microsoft, Oracle, Salesforce and Google, were sales and go-to-market strategies. I worked with the biggest companies in mixed industries, and key application partner ISVs like SAP and Oracle, as well as implementation partners like Deloitte and Accenture, and regional partners.

{audio:0:10:51} How would you characterize the roles of AWS, Microsoft and Google in the IaaS infrastructure market, and what role does Snowflake play? 

{audio:0:11:19} We always positioned Microsoft as being the best end-to-end enterprise for your business, but they lacked one key mission critical application which was ERP or human capital. They bought Dynamics to get into the SaaS play. When a company needs customer experience, ERP or human capital, they will not call Microsoft first.

{audio:0:12:06} Dynamics was key for global Fortune 1,000, but Fortune 500 would call SAP or Oracle ERP. Many decisions are influenced by analyst reports from Gartner and Forrester. Oracle had the most end-to-end SKUs with applications like Oracle Financials, Oracle Supply Chain Management, ERP and Oracle Human Capital Management. They are like Microsoft plus SAP, and are more than simply databases, but they are behind in the cloud journey.

{audio:0:13:07} They cannot catch up in the IaaS space but are leaders in application and SaaS. SAP are struggling as they still have to convert or SaaSify their products to be cloud ready, and need to partner with Microsoft or Google to help them. Google is a consumer search company and have strong intelligence in machine learning and AI, and are a distant third.

{audio:0:14:03} I moved over to Google because I worked with Thomas Kurian as an executive sponsor when I was at Oracle. I don't know if the media knew this, but ‘/T’homas had a mindset that if we partner, we can grow our business better, but Larry Ellison wanted to be exclusive and develop our own cloud instead of partnering with AWS and Microsoft. Snowflake's business model is about partnerships. They are multi-cloud, and even though they sell their own infrastructure services which are serverless, they can still run on AWS, Azure or GCP. That is key because most large companies want to be multi-cloud, just like if you were running Windows, you wouldn't put all your applications on Windows, Linux, Unix or mainframe. It is about mitigating that risk. Google developed Anthos; are you familiar with that product?

{audio:0:15:44} Anthos was a true Kubernetes orchestration engine for multi-cloud. They were pitching to customers who run a hybrid cloud model that they would give them the best tools to seamlessly transition applications between GCP, AWS and Azure. Until we are fully in the cloud, you have to think of all the applications which cannot be moved to the cloud yet, because many of those need a total rewrite of their code. They were built on a three-tier model, not into micro services, and moving them takes a huge amount of work. I manage a large retailer in the US where most of the retail is still operated by IBM mainframes or AS400s. That runs key supply chain logistics and point of sales machines, which are critical systems. We tried for six months to migrate a key supply chain application, but we failed because the code was built in early 80s. Google bought Cornerstone to help with mainframe modernization, but realized the cost of rewriting an application was more than their current support model.

{audio:0:17:40} How would you characterize AWS? 

{audio:0:17:46} AWS started in 2006 which was years ahead, and have almost 270 services. They are by far the leader with the deepest industry verticalization and the strongest set of tools. They have the most Fortune 500 customers and Amazon's earnings are based on AWS growth.

{audio:0:18:27} Do you have an idea of the customer numbers of these three clouds? 

{audio:0:18:31} AWS have 30,000 big customers, but many companies have a dedicated team to help startups, which is where they focus. Those companies you and I have never heard of are still running AWS or those hyperscalers. 

{audio:0:19:04} AWS says they have millions of customers. 

{audio:0:19:08} They have millions but I focus on the Fortune.

{audio:0:19:12} Are you saying in the big accounts there are 30,000 AWS customers? 

{audio:0:19:15} 99% of Fortune 1,000 companies have AWS in one way, shape or form. These are companies with over a billion dollars of revenue, and many of them started with applications which were not considered mission critical. Google has an advantage because many customers are already engaged with them on the ads and revenue search side. An oil company doesn't sell to the public consumer but they still use Google Analytics.

{audio:0:20:12} Is Snowflake Infrastructure-as-a-Service or Platform-as-a-Service?

{audio:0:20:30} They are considered data warehouse-as-a-service, but they have their own niche. Platform-as-a-Service provides someone with tools to help them develop applications, but Snowflake doesn't have a rich set of tools, they do more database management, but you could see them doing that in the future, taking the same business model Salesforce did.

{audio:0:21:17} When Salesforce came out, they launched CRM and developed tools to make it intuitive for users to do configuration. The key difference is you don't need developers to develop applications; you only need to configure them with user friendly widgets. Snowflake's advantage is they have a very intuitive user experience. It is much easier to manage a Snowflake data warehouse than one from Redshift, BigQuery or Azure. You and I may not know how to do a SQL injection query or the JSON languages, but Snowflake makes it easier. You can see them adding more of that capability, so a business user in finance or sales can run queries and develop applications on top. Then it will become a true Platform-as-a-Service.

{audio:0:22:24} Are they going in that direction? 

{audio:0:22:28} With CRM, did you mean the link to Salesforce being that they started with one application then extended their offering? Is that the parallel with Snowflake? 

{audio:0:22:57} Salesforce started with one key application called CRM, then added marketing cloud and acquired other companies, but then they added Platform-as-a-Service which is the Salesforce platform, which allowed business users to add more functionality to their CRM or the marketing cloud. Snowflake will be doing that. 

{audio:0:23:34} Redshift has SageMaker Studio which is a set of rich tools business users can use to develop ML models without learning Python, and BigQuery does the same thing.

{audio:0:23:55} Why would a Fortune 500 choose Snowflake over Redshift, Synapse or BigQuery? I know BigQuery is more competitive and co-sells less than the other two? 

{audio:0:24:31} The decision criteria for data warehousing involves performance, scalability, security, integration with existing systems and how they implement ELT and ETL data pipes. Pricing is key but performance is based on the type of workload. Snowflake's workloads are good when you have steady continuous usage patterns; you can easily scale up or down and their pricing is on demand and you can buy them up front, which makes it easy to budget. BigQuery is designed for spiky workloads like scattered month-end financial report queries or sales during the holiday season where you need to run complex queries for sales intelligence. Their pricing model is more favorable because it is based on number of queries. Redshift is for constant computing, data and streaming, which is why NASDAQ is a big customer of theirs. Data architecture is key but many customers like Snowflake because they don't lock them into one vendor, and customers hate vendor lock in.

{audio:0:26:50} But they are with Snowflake? 

{audio:0:26:51} Yes, but they can still run it off different clouds, which is why it's broader. Microsoft have a SQL database and recently changed their licensing agreement where, if you want to run SQL on AWS, you have to pay a taxation. Customers who ported many of their SQL applications to AWS five years ago are now paying a taxation unless they port it back. Customers want the flexibility that if AWS decides to charge us more or gets out of this business, they can move over to Azure or GCP. Business users like Snowflake's intuitive interface and prefer that over calling IT to run queries for them, they want to do it themselves. Snowflake offers a greater flexibility which is why they will never go out of business. They will grow exponentially by leveraging the back-end infrastructure of AWS and GCP for load balancing, geo relocation and infrastructure like the CDNs and pipes in the background.

{audio:0:28:34} You are charged based on consumption, which can be number of queries or storage; do you expect Snowflake to grow in a similar fashion as clouds have?

{audio:0:29:01} Do you mean grow as big as they are? 

{audio:0:29:03} If they add more services beyond data warehouse. They are only competing with Redshift, BigQuery and other companies like Teradata, but they will never grow their own compute engine like GCP or EC2, or get all the other databases. This is only data warehouse, what about true storage databases like Amazon Dynamo, S3 or others?

{audio:0:29:43} Could Snowflake provide this one day? 

{audio:0:29:44} They could but they would do better developing their own AI and ML tools. 

{audio:0:29:57} Why could Snowflake not evolve to become a fourth hyperscaler without the infrastructure, like a cloud on top of clouds? 

{audio:0:30:26} You could do a cloud on top of cloud with the abstraction layer, but if you want to be a true cloud provider, you need to build the data centers infrastructure, and one of those is $2 billion. Do they have the capital to build out infrastructure? 

{audio:0:30:43} One data center is $2 billion? 

{audio:0:30:44} Yes, $2 billion to build a data center to host these type of clouds. Google can get away with it because they have the infrastructure for their Search and YouTube. So can AWS with their ecommerce and Microsoft who focus on enterprise computing. Snowflake doesn't have the capital to do that, nor would they want to. I could see a scenario where Snowflake gets acquired by one of these hyperscalers to get their IP or make it less competitive.

{audio:0:31:43} It's impossible to know what share of AWS revenue comes from Redshift; do you have an idea what share of revenue comes from these types of products?

{audio:0:32:04} No, they don't break it down.

{audio:0:32:07} Would it be possible for Snowflake to expand their use cases and workloads to become a comprehensive offering for a developer to develop on Snowflake, and why do they not have any data centers? 

{audio:0:32:33} Were you saying why wouldn't they have more data centers?

{audio:0:32:36} The way you talked about it before with them being multi-cloud, basically Snowflake is a layer above the clouds, where they can commoditize and offer their end customers the best price for any given service; could customers shift to where it's most attractive to have this service?

{audio:0:33:02} Yes, similar to a reseller model? 

{audio:0:33:06} Yes, that's where my question was coming from. 

{audio:0:33:11} In those reseller models, you have to ensure you have terms with the companies. When I was at Google, we valued Snowflake as a partner and they were in our marketplace, but upper management decided they were a direct competitor and we were disincentivized to go with them. Azure likes Snowflake because they see them as best in class, and as long as customers expand the Snowflake footprint, the underlying compute and storage will be Azure. 

{audio:0:34:23} I know AWS thinks the same on the partnership; why do these two see it that way and Google not?

{audio:0:34:31} Google thought they could develop their own much better integration with them. Google hire the smartest and probably have the most PhDs and patents. I think it was the wrong decision because Snowflake is definitely gaining more momentum than BigQuery. Thomas Kurian thought they had better control and development with their products, and that they could do a better job in the data warehouse space.

{audio:0:35:32} Does AWS benefit from more workloads on Snowflake in AWS? 

{audio:0:35:43} Once a Fortune 500 company chooses Snowflake, how likely are they to switch or churn away and replace it? AWS is not competitive with Redshift but I have heard they are trying to re-architect that to be competitive with Snowflake. Can AWS succeed and if they do, would existing Snowflake customers churn?

{audio:0:36:47} If a customer wants to make a technology change, something has to be broken or painful with the current model. Most customers change because of performance – especially with this being the analytics for the business – and price. If AWS offers Redshift at half the cost, and they say they have done this with large customers at both Google and Microsoft, and AWS says if you sign a multi-year agreement, they will throw in this many more instances or credits, which you can use to run a data warehouse like Redshift or Dynamo, or even S3, depending on different workloads. 

{audio:0:37:49} If the customer is paying Snowflake $300,000 a year and could reduce that cost by 50%, that might be worth it, especially if they offer tools to help easily migrate. If performance decreased, business users would complain that a query which used to take seconds, now takes minutes, but I don't think Snowflake performance will be lacking, because they claim to have the best performance and do. They always have a strong loyalty to business users because of intuitive interface, so unless they change that, it will be hard for the others to move that workload over. For the same reason, companies still run SAP, even though that architecture has been around for 20 years. They are used to it and trust it, but when they move over their analytics workload from SAP HANA to Snowflake, they realize they can run it faster and easier. 

{audio:0:39:16} When the finance department does their budget, they realize the total cost of ownership of Snowflake is 50% cheaper than SAP because HANA is still on premise and their support fees are exponentially more expensive as their systems age. Things will always change and we've seen hyperscalers offer deeper discounts because of their scale and volume. They could also offer engineering resources to Redshift of BigQuery. They could buy Snowflake or Teradata for their IP to put it into their BigQuery so that BigQuery can do a better job handling real time workloads. They could always change their pricing models because at BigQuery they also sell ML, compute, storage and network. Data warehouse is such a key analytics business function that once business users are familiar, it could take five to 10 years before they switch.

{audio:0:40:42} Snowflake has onboarded a quarter of the Fortune 2000 and half the Fortune 500; will these companies expand their data warehouse with Snowflake?

{audio:0:40:57} The go-to-market strategy of Snowflake is to first take the tier 2 and 3 applications like departmental apps, then once customers like it, they organically sell that internally. Next, they target core finance and supply chain, and finally the SAP ERP analytics database on HANA. One of the largest US energy companies is moving their SAP HANA. They had spent $500 million over five years and started out with a tier 2 and 3 application.

{audio:0:42:01} How does Snowflake improve its performance, because you said it's in their hands to mess it up. If they continue to drive down price and improve performance, which is what they communicate, the customers won't change. 

{audio:0:42:35} The high-level architecture of running queries needs to cater for billions of transactions in a day, so to run faster queries you need to develop smarter algorithms, ELT and ETL tools. It is a serverless architecture which is code based so they don't add more machines. 

{audio:0:43:20} Is the competitive driver scale because it is driven by usage? 

{audio:0:43:38} No. To increase the performance of these queries, you need smarter engineers to write smarter algorithms, because the table is getting bigger, so you do that through code. Increasing volumes lowers prices but performance is through the storage and network spec, and you can choose serverless or be more self-managed. 

{audio:0:44:32} Have you worked with Snowflake on a project? 

{audio:0:44:37} When I was at Google, we worked with Snowflake for retail data management, but then we changed direction and were told not to work with Snowflake and to push BigQuery. 

{audio:0:44:56} Could you characterize Snowflake using adjectives on their good/bad sides? 

{audio:0:45:09} Snowflake is super good at what they do in the data warehouse-as-a-Service. If you need a heart transplant, you see a heart transplant doctor, not a cardiologist or GP. Similarly, if you want more than to simply run data warehouse analytics, you would use one of the enterprise players. Since data analytics is key in every department, Snowflake's can have conversations with finance, HR, supply chain, sales, manufacturing and product management. When they were smaller, they relied on the hyperscalers to get them in front of customers, but now they have gained traction and are the major leader for the past three or four years. In the early days people didn't know what Salesforce was because they knew Peoplesoft and Oracle, but now they know Salesforce. They hired people from Microsoft, Oracle and Google who knew how to sell to enterprises. Snowflake are still immature in the sales processes and have not reached operational excellence like Oracle or Microsoft have, but they will get there, which is why they are hiring people like myself, to up level their sales maturity and motions.

{audio:0:47:14} Who are Snowflake's competitors?

{audio:0:47:23} Other than the hyperscalers, Teradata is the other one I see a lot of. 

{audio:0:47:29} Databricks is much smaller and I was handling more of the Fortune 100. Databricks is very popular but not as big as Snowflake, but I do see them in smaller accounts. If Databricks does a great job managing startups, then five years from now these startups will be Fortune 100 and they will grow their footprint exponentially.

{audio:0:48:22} We discussed Infrastructure-as-a-Service, Platform-as-a-Service and Software-as-a-Service, does the fact that Microsoft and Google are across the entire stack give them a competitive advantage, whereas Amazon is focused on Infrastructure-as-a-Service and Platform-as-a-Service?

{audio:0:49:32} In Infrastructure, Application and Platform-as-a-Service, Amazon is making a big push. For ML tools, all they have is SageMaker, so they are still Platform-as-a-Service. If you look at the code people are writing, it is no longer Microsoft .net, it's JSON, Python or Ruby. Their spearhead will be through ML and AI, which is gaining the most interest from customers.

{audio:0:50:20} How do Microsoft and Amazon compete with each other? I know Microsoft often bundle Office 365 licenses in these deals.

{audio:0:50:39} Microsoft offers bundles to their large customers through enterprise agreements. A typical enterprise has on-premise licensing of Office and Windows, but they can also have Office 365 cloud licensing. They also have SQL licensing, so Microsoft offers them credits if they move a workload from on premise to the cloud. License mobility is big because if you move this license to our cloud, you don't have to pay it, but if you move it to Google or AWS, they will charge you to run those instances in their cloud. Microsoft has a huge license advantage, which some customers like but others feel locked in, which might be a liability.

{audio:0:51:42} The Infrastructure-as-a-Service market is developing, Amazon have 60% market share, Microsoft is next and Google is a distant third, which is what you said from seeing the data. I'm trying to understand why is this constant and whether Microsoft could overcome Amazon through SaaS. Most people think the infrastructure layer is a commodity and the value accrues to the SaaS or PaaS piece, do you agree?

{audio:0:52:35} From a customer standpoint, infrastructure is a commodity and SaaS is the application. They get value from apps which is why iPhone is so popular; it's not the OS kernel underneath. From a company revenue perspective, IaaS is unlimited whereas SaaS is capped by number of users. That is why Microsoft are investing in Azure versus trying to be like a Workday or Salesforce. It will help them grow their cloud base. The Dynamics and Microsoft Cloud business will be a way smaller SaaS business than IaaS.

{audio:0:53:30} Why is the infrastructure piece so profitable for Amazon? 

{audio:0:53:34} 80% to 90% of current workloads still need to be migrated which is consumption, but every new application a business needs will be developed in the cloud because of agility, time to market, security and flexibility. Morgan Stanley or Goldman Sachs need fraud detection software developed in two weeks, not six months, so they will build it in the cloud, and they will use the most trusted cloud provider to do that. The CTO or VP of JP Morgan was a keynote speaker at re:Invent and spoke about partnering with AWS because of their credibility and maturity in the marketplace. NASDAQ partner with Google and AWS as they don't want their eggs in one basket, but there will always be a primary, secondary and tertiary.

{audio:0:55:15} Do you have a Snowflake customer whose AWS spend you know?

{audio:0:55:29} Yes, one of the largest oil companies in the world spent $600,000 a year on Azure, but they have a run rate of $1 million. Azure is still only 10% to 15% of their overall Microsoft spend, but it's growing quickly as they move more applications to Azure.

Snowflake Positioning vs AWS, GCP, Azure

Former Director of Google Cloud

inpractise.com/articles/snowflake-positioning-vs-aws-gcp-azure

Executive Bio

Former Director of Google Cloud

Interview Transcript

Related Content

Azure, GCP, BigQuery, Snowflake & Cloud Competition

Former Director of Google Cloud

AWS, GCP, Azure, Oracle & Cloud Hyperscaler Competition

Former Director of Google Cloud

Investor Dialogue: Amazon Web Services

Snowflake, Synapse, and Redshift: Customer Angle

Former Director of Technology at Capital One

Partner Interview